MoST: Multi-modality Scene Tokenization for Motion Prediction.
Norman MuJingwei JiZhenpei YangNate HaradaHaotian TangKan ChenCharles R. QiRunzhou GeKratarth GoelZoey YangScott EttingerRami Al-RfouDragomir AnguelovYin ZhouPublished in: CoRR (2024)
Keyphrases
- multi modality
- motion prediction
- multi modal
- short term
- medical images
- information theoretic
- motion estimation
- kalman filter
- image registration
- mutual information
- d scene
- video sequences
- three dimensional
- single image
- long term
- imaging modalities
- motion vectors
- video coding
- input image
- moving objects
- medical imaging
- inter frame
- deformation field